• Quantitative Bio-Science
• /
• v.37 no.2
• /
• pp.133-141
• /
• 2018

Graphical lasso is one of the most popular methods to estimate a sparse precision matrix, which is an inverse of a covariance matrix. The objective function of graphical lasso imposes an ${\ell}_1$-penalty on the (vectorized) precision matrix, where a tuning parameter controls the strength of the penalization. The selection of the tuning parameter is practically and theoretically important since the performance of the estimation depends on an appropriate choice of tuning parameter. While information criteria (e.g. AIC, BIC, or extended BIC) have been widely used, they require an asymptotically unbiased estimator to select optimal tuning parameter. Thus, the biasedness of the ${\ell}_1$-regularized estimate in the graphical lasso may lead to a suboptimal tuning. In this paper, we propose a two-staged bias-correction procedure for the graphical lasso, where the first stage runs the usual graphical lasso and the second stage reruns the procedure with an additional constraint that zero estimates at the first stage remain zero. Our simulation and real data example show that the proposed bias correction improved on both edge recovery and estimation error compared to the single-staged graphical lasso.

• Journal of Power Electronics
• /
• v.19 no.3
• /
• pp.655-664
• /
• 2019

In order to analyze the nonlinear phenomena of the bifurcation and chaos caused by the switching of nonlinear switching devices in inductively coupled power transfer (ICPT) systems, a Jacobian matrix model, based on discrete mapping numerical modeling, is established to judge the system stability of the periodic closed orbit and to study the nonlinear behavior of Hopf bifurcation in a system under full resonance. The general flow of the parameter design, based on the stability principle for ICPT systems, is proposed to avoid the chaos and bifurcation phenomena caused by unreasonable parameter selection. Firstly, based on the state equation of SS-type compensation, a three-dimensional bifurcation diagram with the coupling coefficient as the bifurcation parameter is established with a numerical simulation to observe the nonlinear phenomena in the system. Then Filippov's method based on a Jacobian matrix model is adopted to deduce the boundary of stable operation and to judge the type of the bifurcation in the system. Then the general flow of the parameter design based on the stability principle for ICPT systems is proposed through the above analysis to realize stable operation under the conditions of weak coupling. Finally, an experimental platform is built to confirm the correctness of the numerical simulation and modeling.

• Journal of the Korean Mathematical Society
• /
• v.36 no.5
• /
• pp.1009-1020
• /
• 1999

The minimum permanent and the set of minimizing matrices over the face of the polytope n of all doubly stochastic matrices of order n determined by any staircase matrix was determined in [4] in terms of some parameter called frame. A staircase matrix can be described very simply as a Ferrers matrix by its row sum vector. In this paper, some simple exposition of the permanent minimization problem over the faces determined by Ferrers matrices of the polytope of n are presented in terms of row sum vectors along with simple proofs.

• Proceedings of the Korean Society for Noise and Vibration Engineering Conference
• /
• 2007.11a
• /
• pp.863-869
• /
• 2007

An improved method based on a normal frequency response function (FRF) is proposed to identify structural parameters such as mass, stiffness and damping matrices directly from the FRFs of a linear mechanical system. The method for estimating structural parameters directly from the measured FRFs of a structure is presented. This paper demonstrates that the characteristic matrices are extracted more accurately by using a weighted equation and eliminating the matrix inverse operation. The method is verified for a four degree-of-freedom lumped parameter system and an eight degree-of-freedom finite element beam. Experimental verification is also performed for a free-free steel beam whose size and physical properties are the same as those of the finite element beam. The results show that the structural parameters, especially the damping matrix, can be estimated more accurately by the proposed method.

• Journal of information and communication convergence engineering
• /
• v.1 no.4
• /
• pp.209-212
• /
• 2003

In this paper, we proposed new speech feature parameter through parts-based feature extraction of speech spectrum using Non-Negative Matrix Factorization (NMF). NMF can effectively reduce dimension for multi-dimensional data through matrix factorization under the non-negativity constraints, and dimensionally reduced data should be presented parts-based features of input data. For speech feature extraction, we applied Mel-scaled filter bank outputs to inputs of NMF, than used outputs of NMF for inputs of speech recognizer. From recognition experiment result, we could confirm that proposed feature parameter is superior in recognition performance than mel frequency cepstral coefficient (MFCC) that is used generally.

• The Transactions of the Korean Institute of Electrical Engineers A
• /
• v.51 no.11
• /
• pp.577-584
• /
• 2002

This paper presents a novel approach based on eigenvalue perturbation of augmented matrix(AMEP) to estimate the eigenvalue for variation of controller parameter. AMEP is a useful tool in the analysis and design of large scale power systems containing many different types of exciters, governors and stabilizers. Also, it can be used to find possible sources of instability and to determine the most sensitivity parameters for low frequency oscillation modes. This paper describes the application results of AMEP algorithm with respect to all controller parameter of KEPCO systems. Simulation results for interarea and local mode show that the proposed AMEP algorithm can be used for turning controller parameter, and verifying system data and linear model.

• Proceedings of the IEEK Conference
• /
• 2009.05a
• /
• pp.67-69
• /
• 2009

This paper deals with the consensus problem for multi-agent linear dynamic systems with parameter uncertainties. All the agents are identical high-order linear systems with parameter uncertainties and their state information is exchanged through a communication network. It is shown that a consensus is achieved if there exists a feasible solution to a set of linear matrix inequalities obtained for a simultaneous stabilization problem for multiple systems. A numerical example is presented to show the effectiveness of the proposed method.

• Honam Mathematical Journal
• /
• v.29 no.1
• /
• pp.75-81
• /
• 2007

The Fisher information matrix plays a significant role in statistical inference in connection with estimation and properties of variance of estimators. In this paper, the parameter space of the normal distribution using its Fisher's matrix is defined. The Riemannian curvature and J-divergence to parameter space are calculated.

• Honam Mathematical Journal
• /
• v.28 no.3
• /
• pp.433-438
• /
• 2006

The Fisher information matrix plays a significant role in statistical inference in connection with estimation and properties of variance of estimators. In this paper, the parameter space of the t-distribution using its Fisher's matrix is defined. The Riemannian and scalar curvatures to parameter space are calculated.

• 제어로봇시스템학회:학술대회논문집
• /
• 2001.10a
• /
• pp.68.5-68
• /
• 2001

This paper investigates the problem of linear parameter-dependent output feedback controllers design for interconnected linear parameter-varying(LPV) plant. By using a parameter-independent common Lyapunov function, sucient conditions for solving the problems are established, which allow us to design linear parameter dependent decentralized controllers in terms of scaled H-infinite control problems for related linear systems without interconnections. The solvability conditions are expressed in terms of finite-dimensional linear matrix inequalities(LMI´s) evaluated at the extreme points of the admissible parameter set.